Thinking Labs, led by Mira Murati, is tackling nondeterminism in large language models, aiming for consistent outputs regardless of server load. Researchers believe current LLMs like ChatGPT and ...